W3: Data Visualization

Announcements

Last Week Today

(Sorry, John Oliver)

Clear Points

implicit subsetting

Glad it was clear!

today was pretty clear, i need to figure out the syntax for diving into specific parts of data frames but I think I need to just play with this.

Keep at it!

Clear Points

How to make a vector, how to select values out from the vector, how to select values from a data frame, … lots of things!

how to use vectors, what they are used for, and how to start thinking about manipulating data frames

Yay!

Muddy Points

implicit subsetting

We will revisit this when we get to data wrangling in a couple weeks. Let it sit and we’ll revisit it.

Muddy Points

Can vectors be thought of as data columns or are they sometimes rows or neither?

In R, the columns of a data.frame are vectors. These three things are equivalent:

  • Column
  • Variable
  • Vector

In general, rows are not vectors in a data.frame. This is because you can have mixed data types across a row.

Muddy Points

When to use $

We use $ to refer to a column within a data.frame. We mostly do it when we need to implicitly subset based on a criteria of that column. One of the joys of the tidyverse is that you don’t have to use it.

Muddy Points

Nothing muddy but i think since vectors are such a crucial concept to learn - live coding with a simpler data set would be easier. Sometimes the medical data can go over my head. […]

Some of the in class exercises were difficult to follow because we could not see the initial vector definition before each operation was applied to it. This made it hard for me to understand […]

Thanks for the feedback. There is a preview of the data we’re using today in a slide.

I have used toy datasets before.

Muddy Points

Not necessarily unclear, but it would be helpful to do a quick rundown of any tips to remember the ordering for how to recall information and also the ordering of arguments for other functions.

For the most part, we’ve been using the order to distinguish our arguments from each other. You can mix the order up by using the argument names. These are all equivalent:

seq(1,10,2)
seq(from=1, to=10, by=2)
seq(by=2, from=1, to=10)

When working with functions, auto-completion is your friend. Try using the key when you start working with a function.

Data Visualization

Penguins Dataset

gt::gt(head(penguins))
species island bill_length_mm bill_depth_mm flipper_length_mm body_mass_g sex year
Adelie Torgersen 39.1 18.7 181 3750 male 2007
Adelie Torgersen 39.5 17.4 186 3800 female 2007
Adelie Torgersen 40.3 18.0 195 3250 female 2007
Adelie Torgersen NA NA NA NA NA 2007
Adelie Torgersen 36.7 19.3 193 3450 female 2007
Adelie Torgersen 39.3 20.6 190 3650 male 2007
  • Note that our dataset has column names
  • In ggplot2, we don’t need to use the $ operator: penguins$species
  • We use the bare column name to refer to it: species
    • bill_depth_mm : numeric
    • bill_length_mm : numeric
    • species : character

{visdat} for Exploratory Data Analysis

library(visdat)
vis_dat(penguins)

Common Plots

One Variable

  • Numeric: histogram
  • Character: bar plots

Two Variables

  • Numeric vs. Numeric: Scatterplot, line plot
  • Numeric vs. Character: Box plot

Why focus on these plots?

We build a plot one part at a time

Data +

Mapping to data +

Geometry

Think about making plots like using recipes from a cookbook: https://r-graphics.org/

One variable plots

Building a Histogram

ggplot(penguins) +

aes(x = bill_length_mm) +

geom_histogram()

Data +

Mapping to data +

Geometry

ggplot(penguins)

ggplot(penguins) +

  • We always start with ggplot()
  • The first argument to ggplot() is the data
  • We add details to the plot with the + (plus sign)

aes():

aes(x = bill_length_mm) +

  • We map data in with the aes() (aesthetic) function
  • x is an aesthetic - it maps data to a visual property
  • In the aes() function, we use bare column names: bill_depth_mm
  • If you want to know what aesthetics to map, look at the geom documentation:

Some aesthetic properties

Aesthetic Description
x x-coordinate on graph
y y-coordinate on graph
color color of point or line
alpha transparency
size size of point or thickness of line
group group that data belongs to

Not all geom_s support every aesthetic properties

geom_histogram()

geom_histogram()

  • All geometries begin with geom_
  • geom_s require specific aesthetics
  • Tells ggplot2 how to arrange data on page
  • When in doubt, look at the documentation:
    • ?geom_histogram

Taking it one part at a time

ggplot(penguins)

Taking it one part at a time

ggplot(penguins) +
  aes(x = bill_length_mm)

Taking it one part at a time

ggplot(penguins) +
  aes(x = bill_length_mm) +
  geom_histogram()

Histogram recap

ggplot(penguins) +

aes(x = bill_length_mm) +

geom_histogram()

Bar plots

Made for categorical data. Bar plots automatically count each group for you, so you only need to provide one variable (axis).

ggplot(penguins) +

aes(x = species) +

geom_bar()

Two Variable Plots

Scatterplot

ggplot(penguins) +

aes(x = bill_length_mm, y = bill_depth_mm) +

geom_point()

Scatterplot (data)

ggplot(penguins)

Scatterplot (aesthetics)

ggplot(penguins) +
  aes(x = bill_length_mm, 
      y=bill_depth_mm)

Scatterplot (geometry)

ggplot(penguins) +
  aes(x = bill_length_mm, 
      y=bill_depth_mm) +
  geom_point()

Note: Where to put aes()

Our code looks like this:

ggplot(penguins) +
  aes(x = bill_length_mm, y=bill_depth_mm) +
  geom_point()

Most ggplot code looks like this:

ggplot(penguins, mapping = aes(x = bill_length_mm, y=bill_depth_mm)) +
  geom_point()

Either is acceptable!

Boxplot

ggplot(penguins) +

aes(x = species, y = bill_depth_mm) +

geom_boxplot()

What about more than two variables?

Three Variables

ggplot(penguins) +

aes(x = bill_length_mm, y = bill_depth_mm, color = species) +

geom_point()

Additions to Basic Plots

And the Rest

Data +

Mapping to data +

Geometry +

Layout Options

Histogram with a plot theme

ggplot(penguins) +

aes(x = bill_length_mm) +

geom_histogram() +

theme_bw()

Histogram with options

ggplot(penguins) +

aes(x = bill_length_mm) +

geom_histogram(bins = 10)

Faceting

Stratify our plot based on another categorical variable

ggplot(penguins) +

aes(x = species, y = bill_depth_mm, color = species) +

geom_boxplot() +

facet_wrap(~island)

Multivariate Scatterplot by facet

ggplot(penguins) +

aes(x = bill_length_mm, y = bill_depth_mm) +

geom_point() + facet_wrap(~species)

Some additional options

ggplot(data = penguins) +

aes(x = bill_length_mm, y = bill_depth_mm, color = species) +

geom_point() +

labs(x = “Bill Length”, y = “Bill Depth”, title = “Comparison of penguin bill length and bill depth across species”)

Layering Geometries

Adding on to a plot

  • You can layer multiple compatible geometries
    • Must share aesthetics
    • Added one layer at a time
ggplot(penguins) +
  aes(x=bill_depth_mm,
      y=bill_length_mm,
      color=species) +
  geom_density_2d() +           # 2d density geom
  geom_point()

geom_tile() + geom_text() = heatmap

Why is this heatmap missing boxes? Hint: look at penguin counts.

Look at the count() function and see if there’s an argument we can set to fill in the missing boxes.

penguin_counts <- count(x=penguins, species, island)
penguin_counts
# A tibble: 5 × 3
  species   island        n
  <fct>     <fct>     <int>
1 Adelie    Biscoe       44
2 Adelie    Dream        56
3 Adelie    Torgersen    52
4 Chinstrap Dream        68
5 Gentoo    Biscoe      124

Missing Values - How to Fix?

A task in your exercise for the week!

ggplot(penguin_counts) +
  aes(x=species, 
      y=island, 
      fill=n) +
  geom_tile() +
  geom_text(aes(label=n), 
            color="white")

esquisse as a helper

Consider the esquisse package to help generate your ggplot code via drag and drop.

library(esquisse)
esquisser(penguins)

For More Practice:

Community Session Next Week

  • Optional
  • Suggest a Topic and Vote in Google Doc

https://bit.ly/ir-w3